Agentic coding models and harnesses have become so effective that a large share of new software is now written by agents. For a good reason, it’s faster, code is at times higher quality than if human written, and there are seemingly no downsides. We thought.
But the majority of new Software feels average, in a sense that it feels a like AI slop, instead of software that was created with craft and care. For all that follows here, it’s important to understand, that AI coding produces impeccable results, if done right. But it all comes down to the process. Years ago we used to start at an idea, sketch it out by hand or on a whiteboard, then refine the user experience and exact optics in Figma, then start development. We used to style every button to perfection, manually set border radii to be concentric and look correctly. Things took time and while something taking longer doesn’t mean that it is better, it does usually mean that more thought has had to be put into.
Today, many teams jump directly from idea to their coding agent (Claude Code, Codex). This quickly gets them an MVP, but not something that excites users. There’s this well known saying that getting the last 20% right takes 80% of the work, and here I would say that getting the last 20% right takes 95% of the work. And it’s these 20% that make the difference between AI slop, and a thought out solution.